I second that Katak255. You need a little 32.768KHz tuning fork to get near 1 part per million. Most multi-MHz crystals are worse than 20PPM and wiggle based on temperature.
30days*24hours*60minutes*60seconds is 2.59 million seconds/month. An error of 1 part per million results in an error of about 2.6 seconds per month. 20PPM is 52 seconds/month while 50 PPM is 130 seconds(over 2 minutes) leading to about a half hour per year.
I’ve written clock code for little micro controllers to tune out the error by skipping or adding a beat here and there and have been able to tune the clock to a few seconds per month using 20 PPM crystals. The error does not wiggle all that much but is mostly a fixed error from the stated frequency if you keep the temperature pretty stable.
30days*24hours*60minutes*60seconds is 2.59 million seconds/month. An error of 1 part per million results in an error of about 2.6 seconds per month. 20PPM is 52 seconds/month while 50 PPM is 130 seconds(over 2 minutes) leading to about a half hour per year.
I’ve written clock code for little micro controllers to tune out the error by skipping or adding a beat here and there and have been able to tune the clock to a few seconds per month using 20 PPM crystals. The error does not wiggle all that much but is mostly a fixed error from the stated frequency if you keep the temperature pretty stable.
Statistics: Posted by trossin — Fri Jun 21, 2024 1:27 am